Commits

Branch selector

 master

User selector

All users

All time

Commit History

Commits on May 7, 2017

• improve style + a bit of perf for ScatterWeightedSum CUDA

  Show description for 6d82523
  

  Aapo Kyrola

  authored and

  facebook-github-bot

  committed
  6d82523

  Copy full SHA for 6d82523

Commits on May 6, 2017

• declare UpdateTimestepBlob inline

  Show description for 9ef196b
  

  Jonathan Weese

  authored and

  facebook-github-bot

  committed
  9ef196b

  Copy full SHA for 9ef196b

• fix enternalinputs handling in AppendNet v2

  Show description for 5d0bc65
  

  Aapo Kyrola

  authored and

  facebook-github-bot

  committed
  5d0bc65

  Copy full SHA for 5d0bc65

Commits on May 5, 2017

• Adding video data layer for caffe2

  Show description for 4e3b159
  

  Du Tran

  authored and

  facebook-github-bot

  committed
  4e3b159

  Copy full SHA for 4e3b159

• fix recompute_blobs_on_backward

  Show description for 90d7562
  

  Aapo Kyrola

  authored and

  facebook-github-bot

  committed
  90d7562

  Copy full SHA for 90d7562

• Fix Split documentation

  Show description for 6114319
  

  Aapo Kyrola

  authored and

  facebook-github-bot

  committed
  6114319

  Copy full SHA for 6114319

• AttentionCell

  Show description for 8530ac3
  

  jhcross

  authored and

  facebook-github-bot

  committed
  8530ac3

  Copy full SHA for 8530ac3

• Optimize memory for RNN attention

  Show description for 8be1311
  

  urikz

  authored and

  facebook-github-bot

  committed
  8be1311

  Copy full SHA for 8be1311

• Add __dir__ so autocomplete in iPython works.

  Show description for 894e93f
  

  Eider Moore

  authored and

  facebook-github-bot

  committed
  894e93f

  Copy full SHA for 894e93f

• Add model saving and loading to resnet50_trainer.py

  Show description for 639549d
  

  HengCV

  authored and

  facebook-github-bot

  committed
  639549d

  Copy full SHA for 639549d

• opsify AccumulateInputGradients

  Show description for dfa303a
  

  Aapo Kyrola

  authored and

  facebook-github-bot

  committed
  dfa303a

  Copy full SHA for dfa303a

• Added tiles and axis as input parameters to Tile Operator

  Show description for 4786c05
  

  Romain Cledat

  authored and

  facebook-github-bot

  committed
  4786c05

  Copy full SHA for 4786c05

• Refactor FTRL optimizer to allow sending Alpha as input blob

  Show description for ebc17cc
  

  Xiaolong Wang

  authored and

  facebook-github-bot

  committed
  ebc17cc

  Copy full SHA for ebc17cc

• LastNWindowCollector

  Show description for b472c2e
  

  kittipatv

  authored and

  facebook-github-bot

  committed
  b472c2e

  Copy full SHA for b472c2e

• Fix typo 'convlution'

  Show description for 0579812
  

  lukeyeager

  authored and

  facebook-github-bot

  committed
  0579812

  Copy full SHA for 0579812

• lstm_benchmark use rnn_cell.LSTM multicell + assertion

  Show description for 4e05074
  

  Aapo Kyrola

  authored and

  facebook-github-bot

  committed
  4e05074

  Copy full SHA for 4e05074

Commits on May 4, 2017

• Generalize LastNWindowCollector

  Show description for 57a9ea1
  

  kittipatv

  authored and

  facebook-github-bot

  committed
  57a9ea1

  Copy full SHA for 57a9ea1

• Include hint about run ID in store handler assertion

  Show description for 804da82
  

  pietern

  authored and

  facebook-github-bot

  committed
  804da82

  Copy full SHA for 804da82

• Remove unused binary fb_run_plan_mpi

  Show description for fa03daa
  

  pietern

  authored and

  facebook-github-bot

  committed
  fa03daa

  Copy full SHA for fa03daa

• Always use halving/doubling allreduce

  Show description for 9d26303
  

  pietern

  authored and

  facebook-github-bot

  committed
  9d26303

  Copy full SHA for 9d26303

• Update gloo dependency

  Show description for e7547f7
  

  pietern

  authored and

  facebook-github-bot

  committed
  e7547f7

  Copy full SHA for e7547f7

• Check GCC version if compiling with CUDA support

  Show description for 00e9ed8
  

  pietern

  authored and

  facebook-github-bot

  committed
  00e9ed8

  Copy full SHA for 00e9ed8

• net_drawer: --input is required

  Show description for d98e2aa
  

  lukeyeager

  authored and

  facebook-github-bot

  committed
  d98e2aa

  Copy full SHA for d98e2aa

• Set optimal number of DAG workers for predictor beam search step-net

  Show description for c4d7f29
  

  James Reed

  authored and

  facebook-github-bot

  committed
  c4d7f29

  Copy full SHA for c4d7f29

Commits on May 3, 2017

• make all tensor.h enforces pass the caller

  Show description for d74aea5
  

  Aapo Kyrola

  authored and

  facebook-github-bot

  committed
  d74aea5

  Copy full SHA for d74aea5

• Add caffe2 operators to mobile: Log, StumpFunc, Div, Sub

  Show description for 193dd8f
  

  Carlos Garcia Jurado Suarez

  authored and

  facebook-github-bot

  committed
  193dd8f

  Copy full SHA for 193dd8f

• Add direct enqueuing to enable RNN input, allow specify batch columns

  Show description for 92474bc
  

  Aapo Kyrola

  authored and

  facebook-github-bot

  committed
  92474bc

  Copy full SHA for 92474bc

• Add ScatterWeightedSum for GPU.

  Show description for 516209d
  

  Pooya Davoodi

  authored and

  facebook-github-bot

  committed
  516209d

  Copy full SHA for 516209d

• MultiRNNCell (Caffe2)

  Show description for 907ded3
  

  jhcross

  authored and

  facebook-github-bot

  committed
  907ded3

  Copy full SHA for 907ded3

• auto -> return type for C++11 support

  Show description for ba4c997
  

  bwasti

  authored and

  facebook-github-bot

  committed
  ba4c997

  Copy full SHA for ba4c997

• Fix pybind11 module name for MPI helpers

  Show description for e410972
  

  pietern

  authored and

  facebook-github-bot

  committed
  e410972

  Copy full SHA for e410972

• Prevent false overwriting of a field

  Show description for 306040f
  

  kittipatv

  authored and

  facebook-github-bot

  committed
  306040f

  Copy full SHA for 306040f

• Inference code for seq2seq model

  Show description for c84ef29
  

  urikz

  authored and

  facebook-github-bot

  committed
  c84ef29

  Copy full SHA for c84ef29

• cuda (sparse) lengths sum

  Show description for e200ed3
  

  harouwu

  authored and

  facebook-github-bot

  committed
  e200ed3

  Copy full SHA for e200ed3

• lstm_benchmark: add warm-up stage, support layers

  Show description for 6351f19
  

  salexspb

  authored and

  facebook-github-bot

  committed
  6351f19

  Copy full SHA for 6351f19

Pagination

Previous

Next